Downloaded from journal.pda.org on July 18, 2015
Filling of High-Concentration Monoclonal Antibody Formulations into Pre-Filled Syringes: Filling Parameter Investigation and Optimization Wendy Shieu, Sarah A. Torhan, Edwin Chan, et al.
PDA J Pharm Sci and Tech 2014, 68 153-163 Access the most recent version at doi:10.5731/pdajpst.2014.00973
Downloaded from journal.pda.org on July 18, 2015
RESEARCH
Filling of High-Concentration Monoclonal Antibody Formulations into Pre-Filled Syringes: Filling Parameter Investigation and Optimization WENDY SHIEU, SARAH A. TORHAN, EDWIN CHAN, AARON HUBBARD, BENSON GIKANGA, OLIVER B. STAUCH, and YUH-FUN MAA* Pharmaceutical Processing and Technology Development, Genentech, a member of the Roche Group, South San Francisco, CA ©PDA, Inc. 2014 ABSTRACT: Syringe filling, especially the filling of high-concentration/viscosity monoclonal antibody formulations, is a complex process that has not been widely published in literature. This study sought to increase the body of knowledge for syringe filling by analyzing and optimizing the filling process from the perspective of a fluid’s physical properties (e.g., viscosity, concentration, surface tension). A bench-top filling unit, comprising a peristaltic pump unit and a filling nozzle integrated with a linear actuator, was utilized; glass nozzles were employed to visualize liquid flow inside the nozzle with a high-speed camera. The desired outcome of process optimization was to establish a clean filling cycle (e.g., absence of splashes, bubbles, and foaming during filling and absence of dripping from the fill nozzle post-fill) and minimize the risk of nozzle clogging during nozzle idle time due to formulation drying at or near the nozzle tip. The key process variables were determined to be nozzle size, airflow around the nozzle tip, pump suck-back (SB)/reversing, fluid viscosity, and protein concentration, while pump velocity, acceleration, and fluid/nozzle interphase properties were determined to be relatively weak parameters. The SB parameter played an especially critical role in nozzle clogging. This study shows that an appropriate combination of optimal SB setting, nozzle size, and airflow conditions could effectively extend nozzle idle time in a large-scale filling facility and environment. KEYWORDS: Prefilled syringe, High-concentration monoclonal antibody, Peristaltic pump, Suck-back, Formulation drying. LAY ABSTRACT: Syringe filling can be considered a well-established manufacturing process and has been implemented by numerous contract manufacturing organizations and biopharmaceutical companies. However, its technical details and associated critical process parameters are rarely published. The information on high-concentration/ viscosity formulation filling is particularly lacking. The purpose of this study is three-fold: (1) to reveal design details of a bench-top syringe filling unit; (2) to identify and optimize critical process parameters; (3) to apply the learning to practical filling operation. The outcomes of this study will benefit scientists and engineers who develop pre-filled syringe products by providing a better understanding of HC formulation filling principles and challenges.
Introduction Over the past decade, pre-filled syringes (PFSs) have become widely used in pharmaceutical industries and are now one of the primary containers of choice for parenteral drug delivery, especially for subcutaneous
*Corresponding Author: Genentech, a member of the Roche Group, 1 DNA Way South San Francisco, CA 94080. Telephone: 650-225-3499; Fax: 650-742-1504; email: [email protected] doi: 10.5731/pdajpst.2014.00973
Vol. 68, No. 2, March–April 2014
and intramuscular administration (1). It is particularly true for high-dose monoclonal antibody (mAb) products, which often require several milligrams of protein per kilogram body weight. The subcutaneous route of administration is preferred because it enables home administration for patients with chronic conditions, but generally restricts injection volume and time. Thus, a high-concentration (HC) mAb formulation (e.g., ⱖ100 mg/mL) is often required (2, 3). The challenge of a HC mAb formulation in the PFS product is well known—its high fluid viscosity may make syringes less desirable to use (e.g., high injection force) and formulation stability is more difficult 153
Downloaded from journal.pda.org on July 18, 2015
to maintain (3–9). The impact of HC and high-viscosity formulations on processing and manufacturing, however, is not as well documented, despite the fact that manufacturing a PFS product is a complex process that includes liquid formulation preparation (thawing, compounding, sterile filtration, etc.), component assembly (syringe, stopper, needle, and needle shield), syringe filling, stopper placement, labeling, and packaging (4, 5). Syringe filling is a complicated, but mature, operation; highly automated filling facilities have been established for many years to fill syringes while meeting strict fill weight specifications at a production rate of ⬎10,000 units/h. Yet, for engineers and scientists who are interested in learning or developing a PFS filling process, publications or disclosures detailing the syringe filling operation are scarce. Particularly lacking is published information regarding filling HC mAb formulations. This study aimed to increase the body of knowledge on HC mAb formulation syringe filling by providing insight into filling process optimization. When filling syringes using a pump mechanism, there are two sets of filling parameters: filling nozzle movement and liquid flow generated by the pump. The motion of the nozzle and the flow of the liquid must be aligned to deliver a clean filling profile, where the nozzle retracts and maintains a constant distance from the liquid level, thereby having no contact with the fluid/liquid (too close) or causing undesired splashing/ foaming/bubbles (too far apart). In addition, liquid dripping at the end of the fill should be minimized to enhance fill weight (volume) accuracy and reduce the risk of wet stoppers. When developing the PFS filling process, process engineers often use a fixed nozzle movement profile and adjust the pump parameters to complete the fill cycle. This is particularly true when the nozzle motion is driven by a mechanical cam with pre-set nozzle movement parameters controlled by a single variable—the main drive speed. Thus, this study focused only on investigating the parameters of the peristaltic pump, which is considered a standard pumping mechanism for biopharmaceutical formulations because of its mild stress and closed-system feature compared with the piston pump (10, 11). Modern peristaltic pumps control and deliver liquid through three variables: acceleration, velocity, and reversing (also called suck-back [SB] or back-suction). Acceleration and velocity determine the rate of the liquid filled into the syringe, while SB offers a particular func154
tion—withdrawing the liquid to minimize dripping at the end of each fill. Sporadic dripping may reduce fill weight accuracy. This study also delineated another function of SB, unique to HC protein formulations, in decreasing the probability of nozzle clogging due to formulation drying at the tip of the nozzle during filling interruptions (or idle time). The impact of SB in relation to several process variables on formulation drying was investigated in order to prolong filling interruption times for fluids of varied viscosity, including a model protein and a mAb formulation. The prolonged interruption time derived from this study may provide increased processing flexibility and may facilitate risk mitigation at filling sites. Materials and Methods All experiments in this study used a bovine serum albumin (BSA, at 370 mg/mL) or a mAb (mAb A at 180 mg/mL). These two molecules were formulated into the same composition containing 0.04% Tween 20. Some experiments also used distilled water containing 0.04% Tween 20 for comparison. These formulations were filled into either 25 mL PETG bottles and/or 1.0 mL long, 27G 1⁄2 inch staked needle syringes using a bench-top filling system. All equipment and materials used in these studies are listed in Table I. Bench-Top Filling System The bench-top filling unit was assembled and tested by Volo Technologies, Inc. (Roseville, CA). This unit (Figure 1) integrates a Flexicon peristaltic pump system with a linear actuator to control nozzle movement. The pump system consists of a peristaltic pump and controller, while the robotic system is composed of a ROBO Cylinder® linear actuator and a Volo-integrated controller. Peristaltic Pump Control and Tubing Arrangement: The liquid formulation was delivered using a Flexicon PD12 peristaltic pump (“a” in Figure 1) whose parameters (velocity, acceleration, SB) were controlled by a Flexicon MC12 control unit (“b” in Figure 1). Two pieces of Sani-Tech® platinum-cured 1.6 mm silicone tubing ran through the pump and were Y-connected to 3.2 mm silicone tubing before and after the pump unit (see Figure 1). This tubing arrangement was maintained throughout the whole study. Robotic Movement Control: The linear actuator (“c” in Figure 1) provides the diving action for the filling nozzle. The nozzle moves up and down per PDA Journal of Pharmaceutical Science and Technology
Downloaded from journal.pda.org on July 18, 2015
Table I Equipment and Materials Used in the Study Equipment Stainless steel nozzles (1.5, 2.0, and 2.5 mm ID; 90 mm long) Glass nozzles (1.8 and 2.4 mm ID; 90 mm long) Pump controller module (micro linear actuator) ROBO Cylinder linear actuator PETG bottle Peristaltic pump and controller High-speed camera Cone and plate rheometer Contact angle meter
Materials Syringes (1.0 mL long 27G 1⁄2 inch staked needle) Platinum-cured silicone tubing (1.6 mm and 3.2 mm ID) Y-connectors and fittings Formulation buffer Bovine serum albumin (BSA) powder
Model/ Supplier/Location INOVA (OPTIMA Pharma GmbH, Germany) Pegasus Industrial Specialties Inc. (Cambridge, Ontario) Watson-Marlow (Ringsted, Denmark)/Flexicon IAI America (Torrence, CA) Thermo Fisher Scientific Inc/Nalgene Nunc International Corporation (Waltham, MA) Watson-Marlow (Ringsted, Denmark)/Flexicon Hindsight GigE, 20/20 Hindsight, Monitoring Technology Corp., (Fairfax, VA) Physica MCR 501with CP20-0.5 measuring cone, Anton Paar GmbH (Austria) Model OCA15, FDS Corp. (Garden City, NY) Supplier/Model Becton Dickinson (Swedesboro, NJ)/Hypak Type 1 glass syringes Saint-Gobain Performance Plastics (Taunton, MA)/Sani-Tech Value Plastics (Fort Collins, CO) Genentech, Inc. (South San Francisco, CA) SeraCare Life Sciences (Milford, MA)
the commands from a Volo controller (“d” in Figure 1), which combines the function of a programmable logic controller and a human–machine interface (HMI). The Volo controller is interfaced with the Flexicon controller to send signals to start and stop
pumping the liquid. To enable diverse nozzle movement patterns (dive-in position, fill position, retraction rate, acceleration/deceleration, etc.), various set points need to be inputted to the Volo HMI and Flexicon unit. Filling Operation and Experiments The whole bench-top filling unit was placed inside a horizontal laminar airflow (LAF) hood at room temperature (20 –23 °C) with airflow at a rate of 0.7 m/s from the back panel of the hood in parallel to the bench and a relative humidity of 40 –50%. The rate of the airflow was not adjustable, but the flow could be turned off.
Figure 1 Bench-top filling system consisting of (a) Flexicon peristaltic pump, (b) pump controller, (c) linear actuator, and (d) Volo controller. Vol. 68, No. 2, March–April 2014
Nozzle Movement Profile (Cycle): The nozzle movement profile was pre-set and fixed (Figure 2) in all experiments. During operation, the nozzle dives 44 mm (measured from the back of the flange) into the syringe (52 mm in total length) and begins delivering the liquid. At the same time, the nozzle retracts with an acceleration of 150 mm/s2 and a constant velocity of 70 mm/s. 155
Downloaded from journal.pda.org on July 18, 2015
sample of 70 – 80 L was loaded onto the lower measuring plate and allowed to come to thermal equilibrium at 20 °C. A solvent trap was used to prevent fluid evaporation during the measurement. The sample dynamic viscosity was measured every 10 s for 2 min using a cone with a 19.98 mm diameter and 0.509 degree angle at a shear rate of 1000/s.
Figure 2 Graphic representation of a fixed-nozzle movement cycle used in the study.
Peristaltic Pump Parameter Setting: The fill rate of the liquid was aligned with the nozzle motion by adjusting acceleration and velocity of the peristaltic pump, as monitored through a high-speed camera. Afterwards, the SB setting, from 0 to 10, of the pump was selected and optimized to minimize liquid dripping at the end of each fill cycle, again with the visual aid of a high-speed camera. Interruption (or Drying) Studies: At the end of the fill, the filling process was interrupted (idled) for a pre-determined duration. After interruption, the fill was resumed and the fill characteristics were monitored using a high-speed camera. Suck-Back (SB) Experiments Using Glass Nozzles: Glass nozzles of different sizes (inner diameters, IDs) were used to visualize (via a high-speed camera) SB action and the flow of the liquid formulation at the tip of the glass nozzle. The SB volume, V, in response to SB settings, ranging from 0 to 10, can be calculated using eq 1 where H is the measured height of the air pocket and d is the nozzle diameter: V[mL] ⫽ H[cm] ⫻ ⫻ (d[cm]/2)2
(1)
Fluid (Formulation) Characterizations Viscosity Measurement: The viscosity of a fluid was measured using a cone and plate rheometer. Each 156
Contact Angle Measurement: Static contact angle of a fluid on glass and stainless steel surfaces was determined using a contact angle meter employing an optical contact angle method called sessile drop. The substrates were cleaned with 2% clean-in-place (CIP) cleaning detergent followed by three rinses with water for injection (WFI). For static contact angle measurement, a photo snapshot is taken once a drop of the fluid (5 L) is dispensed from the syringe and laid on a clean substrate surface (glass slides or stainless steel coupons). The angle between the baseline of the drop and the tangent at the drop boundary is measured on both sides. The complete measurement was obtained by averaging the two numbers. At least five readings were recorded for each sample. Results and Discussion Fluid Characterizations Two important fluid properties, viscosity and contact angle, were measured to determine their impact on filling characteristics and formulation drying patterns. The results are summarized in Table II for three fluids: water containing 0.04% Tween 20, 18% mAb A (buffered), and 370 mg/mL BSA (buffered). These fluids covered a broad range of viscosity, 1 to 25 cP (at 20 °C). Contact angles of a fluid on a substrate represent the interactions of the fluid and the substrate surface; low contact angles suggest hydrophilic interactions, while high contact angles demonstrate hydrophobic behaviors of the interface. Two substrate materials, glass and stainless steel, were used to represent the nozzle materials used in this study. As summarized in Table II, the glass surface is more hydrophilic than the stainless steel surface for all three fluids; the 370 mg/mL BSA solution is the most hydrophobic on both glass and stainless steel surfaces. The impact of viscosity and hydrophilicity versus hydrophobicity on fluid flow behaviors inside and at the tip of the nozzle, as well as fluid drying properties, was assessed and is discussed later. PDA Journal of Pharmaceutical Science and Technology
Downloaded from journal.pda.org on July 18, 2015
Table II Summary of Fluid Characterizations and Peristaltic Pump Parameters with Optimal Filling Profile (No Dripping) of Liquid Formulations from Stainless Steel Nozzles of Different Sizes Characterization/Parameter
Water (with 0.04% Tween 20)
mAb A (180 mg/mL)
BSA (370 mg/mL)
1 53.6 ⫾ 1.0 32.6 ⫾ 1.5 200 4
9 75.2 ⫾ 1.8 25.4 ⫾ 1.1 200 50
25 86.1 ⫾ 1.8 40.3 ⫾ 2.4 200 125
Viscosity at 20 °C (cP) Contact angle on stainless steel surface (°) Contact angle on glass surface (°) Pump velocity (rpm) Pump acceleration (no unit) SB setting 1.5 mm stainless steel nozzle 2.0 mm stainless steel nozzle 2.5 mm stainless steel nozzle
Control of Peristaltic Pump Parameters for Filling Characteristic Optimization Initially, all experiments were performed with stainless steel nozzles. With a fixed nozzle movement profile (Figure 2), pump parameters (i.e., velocity, acceleration, and SB) were visually observed through a high-speed camera to optimize filling characteristics, which included maintaining a constant distance between the fluid level and the nozzle tip and minimizing splashing, foaming, and dripping. Pump velocity and acceleration are expected to dominate the rate of liquid flow, thereby affecting the relative position of the fluid and the nozzle. Splashing and foaming may occur when the distance between the fluid level and the nozzle tip is too far. Splashing at the end of the filling cycle is a particular concern because it may result in wetting the stopper after stopper placement. A wet stopper is a critical product defect and will result in filled syringe rejection. SB does not affect solution flow rate, but it exerts a reversing action at the end of each fill cycle to modulate fluid dripping. Dripping should be prevented or minimized because undesired dripping may decrease fill weight accuracy. Additionally, dripping while the nozzle is moving out of the syringe barrel may cause splashing and/or leave fluid drop(s) on the barrel, potentially leading to wet stoppers. Optimal pump parameters that minimized issues such as splashing, foaming, and dripping for the three fluids evaluated in this study are listed in Table II. It was possible to maintain a pump velocity of 200 rpm for all three fluids, but pump acceleration had to be increased with increasing fluid viscosity (4, 50, and 125 Vol. 68, No. 2, March–April 2014
1 1 1
5 2 2
10 3 2
for water, 18% mAb A, and 370 mg/mL BSA, respectively) to align with nozzle motion. The reason that pump acceleration had to be increased in relation to increased fluid viscosity could be described by the Hagen-Poiseuille equation (eq 2): Q ⫽ R4⌬P/8 L
(2)
For a fluid flowing through a pipe—in this case, the filling nozzle—the volumetric flow rate is the function of pressure drop (⌬P), fluid viscosity (), tube radius (R), and pipe length (L). For a given pipe/tubing (fixed R and L) and a nozzle motion profile at a fixed Q, filling a fluid of higher viscosity must be countered with an increasing pressure drop, which could be achieved with a higher acceleration by the pump (i.e., ⌬PA ⫽ ma, where m is the mass of the fluid, a is acceleration, and A is the surface area where the pressure is applied). SB setting (0 to10) also had to be increased with increasing fluid viscosity to minimize dripping, but it could be decreased using a larger nozzle. These observations may again be explained by the HagenPoiseuille equation (eq 2). A greater suction force (in the direction opposite to the liquid fill) is required to pull a more viscous fluid; thus, a higher SB setting is required for more viscous fluids. However, the required suction force for a larger nozzle is lower; thus, a lower SB setting can be exerted to reduce the fluid’s dripping tendency. The function of SB turned out to be more complex and was found to affect another key filling process perfor157
Downloaded from journal.pda.org on July 18, 2015
Table III Summary of Drying Times for the mAb A Formulation Due to Interruption of Filling from Glass Nozzles as a Function of Suck-Back (SB) Settings Drying Time (min) 1.8 mm Glass Nozzle SB Setting
0.7 m/s Airflow
2.4 mm Glass Nozzle 0.7 m/s Airflow
0 m/s Airflow
0 ⬍10 ⬍15 ⬃20 1 ⬍10 ⬃15 ⬃30 2 30–40 ⬃90 ⬎120* 3 ⬍10 ⬃15 ⬃30 4 ⬍10 ⬃15 ⬃45 *No drying was observed before the end of the experiment.
mance indicator—formulation drying at the fill nozzle tip during a fill interruption. Observation and Effect of Formulation Drying During filling operations, many different type of issues (e.g., misalignments and/or malfunctions of the robotic actions, particles) may trigger filling operation interruptions. During these interruptions, the nozzles remain idle for uncertain durations. Nozzle clogging has been observed at the end of a fill interruption due to the drying of the fluid at or close to the tip of the nozzle, resulting in complete or partial nozzle blockage after the fill was resumed. This drying-induced nozzle blockage is unique to HC and high-viscosity formulations containing high-molecular-weight species (polymer, proteins, mAbs, etc.) Water evaporation from such a fluid rapidly establishes a viscous film at the drying front that can easily become elastic, thicken, or solidify. To fully understand the effect of SB setting and performance, a series of experiments was performed in glass nozzles and visualized via a high-speed camera. These experiments were designed to evaluate the effect of interruption time on formulation drying as a function of SB settings, nozzle size (1.8 mm and 2.4 mm ID), and airflow rate (0 and 0.7 m/s). As summarized in Table III, nozzle clogging took place, and in many cases it occurred within 15 min of idle time. All experiments were performed under ambient conditions 158
Figure 3 Photographic images of (a) dried residual ring forming at the inner nozzle tip and growing toward the center of the nozzle, (b) thinning flow pattern of 370 mg/mL BSA formulation after 30 min idle time, (c) graphic representation of dried ring forming for a small vs large nozzle.
(20 –23 °C and 40 –50% relative humidity) in a LAF hood, where air was either turned off (without airflow) or on (with an airflow rate of 0.7 m/s). Without airflow, it took almost twice the time for the mAb A formulation to clog the nozzle than it did under the high-airflow rate, suggesting that airflow near the nozzle tip is an important parameter affecting formulation drying. It is important to note, however, that the high rate of parallel airflow experienced by the filling nozzle in the LAF may represent a worst-case environment compared with an airflow-controlled manufacturing fill line. Table III also shows that it took longer for the larger nozzle (2.4 mm) to clog compared with the smaller one (1.8 mm). Although the rate of water evaporation per unit cross-section area should be the same regardless of nozzle size, the observation of the dried residual begins along the inner surface near the nozzle tip to form a dried ring, which grew toward the center of the nozzle (Figure 3a). The formation of an inner ring PDA Journal of Pharmaceutical Science and Technology
Downloaded from journal.pda.org on July 18, 2015
was further verified by the narrowing of the liquid flow stream (Figure 3b) after the 370 mg/mL BSA formulation was idled for 30 min. The observation of dried residual formation along the inner perimeter of the nozzle tip can be related to the observation of formation of coffee stains from dried liquid drops, which is attributed to capillary flow (12). There is a prerequisite for this phenomenon to occur—the pinning of the liquid–substrate contact line. As the edge (contact line) of the coffee drop is pinned (fixed), water evaporating from the edge must be replenished by liquid from the interior via capillary flow. In the case of liquid SB inside a filling nozzle, the edge of the fluid meniscus is pinned to the nozzle inner surface where water evaporation will easily leave a dried ring. The liquid from the interior will flow to the edge to replenish water. As this process continues, the dried ring will keep growing toward the center of the nozzle to leave a smaller opening or eventually clog the nozzle. The rate of dried ring growth is comparable between the small and large nozzles and has a more significant impact on the smaller nozzle, which will clog faster than the large nozzle (Figure 3c). SB setting was observed to play an important role on the rate of formulation drying; an interruption time could be substantially extended by optimizing SB setting (i.e., a setting of 2 for the mAb A formulation, in both nozzle sizes). The effect of SB setting on formulation drying can be explained by visual observations of the physical behavior of the formulation in glass nozzles. When the mAb A formulation was sucked into a 2.4 mm glass nozzle, it displayed a unique pattern to each SB setting (0 to 4) as shown in Figure 4a. A liquid drop hung at the tip of the nozzle at a SB setting 0 (without SB). At SB setting 1, the liquid drop was withdrawn into the nozzle to show a slight concave meniscus surface at the very tip of the nozzle. The fluid receded farther into the nozzle as the SB setting was increased to 2, 3, and 4. However, beginning at SB setting 3, an air pocket (“ap” in Figure 4b) was observed to be enclosed by a liquid plug (“lp” in Figure 4b) at the nozzle tip. The size of the air pocket and the liquid plug grew larger when the SB setting was increased to 4. When testing out interruption times at different SB settings for mAb A formulation in glass nozzles, SB setting 2 was observed to have the longest interruption time with no nozzle clogging (Table III). Using the glass nozzles, it was possible to visually observe that there was neither a drop hanging from the nozzle nor a liquid plug at the inner tip of the Vol. 68, No. 2, March–April 2014
Figure 4 Photographic images of mAb A formulation sucked back into a 2.4 mm glass nozzle (a) corresponding to suck-back (SB) settings 0 – 4 under the pump setting of velocity at 200 rpm and acceleration at 50; (b) a typical SB pattern showing an air pocket (“ap”) and a liquid plug (“lp”) at the tip of the nozzle.
nozzle at SB setting 2, which was unlike the physical behavior observed with the other SB settings. From these visual observations, a hypothesis can be made that correlates SB setting to nozzle clogging. Water can evaporate quickly from the liquid drop hanging outside the nozzle tip, and the evaporation rate can be slowed down if the fluid is pulled into the nozzle because of reduced airflow and humidity concentration gradient. Intuitively, the farther the fluid retreats, the slower the drying rate. However, like draining a liquid from a container surface, quick fluid flow/removal is normally not complete and can leave a thin film along the nozzle’s inner surface. Later, this fluid film can migrate down to accumulate at the nozzle tip to form a liquid plug which, like the liquid drop at the nozzle tip, can dry quickly and clog the nozzle. Thus, SB setting is a critical parameter that affects formulation drying/nozzle clogging, and it needs to be optimized to minimize liquid dripping or 159
Downloaded from journal.pda.org on July 18, 2015
Figure 5 Suck-back (SB) patterns for (a) 1.8 mm and (b) 2.4 mm glass nozzles in relation to SB settings of 0 –10.
the formation of a liquid plug at the nozzle tip in order to effectively extend the interruption times, particularly in filling a HC or high-viscosity fluid. The SB setting is likely related to the degree of pump rotation and is dependent on the type of the peristaltic pump used. However, despite its importance, the control of SB and the relationship between SB setting and volume are rarely specified in pump manufacturers’ user manuals, and vendors hesitate to provide any information. To generate such data, glass nozzles can serve as a valuable tool, allowing visual observations and direct measurement to quantify and assess SB performance. Control of SB Performance Having determined that SB setting is a critical parameter affecting formulation drying, additional experiments were performed to assess the effect of other fill parameters on SB. Effect of Viscosity and Nozzle Size: SB was visualized as shown in Figure 5, which portrays a linear relationship between the height of both the air pocket and liquid plug and the SB setting for two different nozzle sizes. This linearity held true in both the 1.8 mm and 2.4 mm nozzles. The SB volume can be calculated using the measured SB height (eq 1). The three solutions listed in Table II were used to assess the effect of viscosity on SB volume. As summarized in Figure 6, viscosity has an obvious impact on SB volume. At the same SB setting, a fluid with higher viscosity is more difficult to be drawn into the nozzle 160
because a higher pressure must be exerted on the tubing to draw more viscous formulations through the pump. Nozzle size appears to play a less significant role, at least for less viscous fluids, such as water and mAb A formulation (⬍10 cP). However, a nozzle size– dependent difference was observed for the more viscous BSA solution (25 cP), which could not be sucked into the 1.8 mm nozzle until the SB setting reached 4. Effect of Pump Parameters—Velocity and Acceleration: The two pump parameters, velocity and acceleration, were evaluated separately to assess their impact on SB volume of the mAb A formulation. In the first experiment, pump velocity was set to 200 rpm and pump acceleration was varied in the range of 25 and 100, represented by the solid lines in Figure 7a. SB volumes were not substantially affected by the level of
Figure 6 Suck-back (SB) volume vs SB setting for 370 mg/mL BSA, mAb A, and water in 1.8 mm and 2.4 mm nozzles. PDA Journal of Pharmaceutical Science and Technology
Downloaded from journal.pda.org on July 18, 2015
0.05 and 0.08 mL for pump velocity ranging between 25 and 400 rpm. Beyond 100 rpm, the bubble volume stayed in an even narrower range of 0.07– 0.08 mL. Because the pump parameters are typically set at ⬎100 rpm, pump velocity was again demonstrated not to be an important variable for bubble volume. Similarly, a fourth experiment was conducted with the 1.8 mm glass nozzle at a fixed SB setting of 3 and a pump velocity of 200 rpm. The results are presented in Figure 7c, where the SB volume varied between 0.05 and 0.11 mL for acceleration in the range of 25–200. Bubble volume increased faster initially (at acceleration of 25–125) but basically fluctuated in a small range (0.09 – 0.11 mL) for acceleration of 125–200. Given that pump acceleration is typically set between 25 and 125 for high-viscosity formulations (using the 1.6 mm silicone tubing), pump acceleration may have a slightly more significant impact on the SB behavior than pump velocity during syringe filling. Overall, pump velocity and acceleration were shown to be relatively weak in affecting SB behavior. Formulation Drying in Metal Nozzles
Figure 7 Suck-back (SB) volume vs SB setting for 370 mg/mL BSA (a) with varying pump velocity (V, dashed lines) and acceleration (A, solid lines) settings; (b) illustration of SB volume at SB setting 3 and acceleration at 75; (c) illustration of SB volume at SB setting 3 and velocity at 200 rpm.
pump acceleration. In the second experiment, pump acceleration was fixed at 50 and pump velocity was varied in the range of 100 and 300 rpm, represented by the dotted lines in Figure 7a. Pump velocity did not affect SB volume significantly either, suggesting that pump velocity did not play a significant role in the SB function of the peristaltic pump. In the third experiment pump velocity was evaluated against air pocket (bubble) formation in the 1.8 mm nozzle at a fixed SB setting of 3 and a pump acceleration of 75. The results are shown in Figure 7b, where the SB volume fluctuated in a narrow range between Vol. 68, No. 2, March–April 2014
Glass nozzles rendered valuable visual information for the understanding of critical pump parameters. However, glass and stainless steel have different surface properties, and it can be argued that observations made with glass nozzles are not relevant to stainless steel nozzles. Thus, an experiment was designed to evaluate whether observations in the glass nozzle can be applied to stainless steel nozzles. Earlier we pointed out that the 370 mg/mL BSA formulation is the most hydrophobic on both glass and stainless steel surfaces, but we also showed a big contact angle difference between the two: 40.3 ⫾ 2.4° on glass and 86.1 ⫾ 1.8° on stainless steel. Thus, this fluid was tested on both glass and stainless steel nozzles of different sizes (see Table IV). The optimal SB setting (i.e., positive SB without forming a liquid plug or liquid drop) in glass nozzles (1.8 and 2.4 mm) were determined to be 4 and 2, respectively. Applying these optimal SB settings, the 370 mg/mL BSA solution still dried much quicker in the 1.8 mm nozzle than in the 2.4 mm nozzle (20 –30 min vs ⬎60 min). For stainless steel nozzles of three different sizes (1.5, 2.0, and 2.5 mm), the formulation drying (nozzle clogging) results again demonstrated the same trend. At a comparable nozzle size (2.4 mm glass nozzle and 2.5 mm stainless steel nozzle), the optimal SB setting and the duration of drying times 161
Downloaded from journal.pda.org on July 18, 2015
Table IV Summary of Drying Studies for 370 mg/mL BSA Formulation Sucked Back into Glass Nozzles or Stainless Steel Nozzles of Different Sizes Stainless Steel Nozzle Study Finding
1.5 mm
2.0 mm
Optimal SB setting 10 3 Drying time for nozzle clogging (min) 5 ⬎60* Contact angle (°) 86.1 ⫾ 1.8 *Drying experiment and nozzle observation did not exceed 60 min.
were identical. For the 2.0 mm stainless steel nozzle, the SB had to be increased to 4 to significantly extend the drying time to ⬎60 min, while the 1.8 mm glass nozzle had an optimal SB setting at 3 and 20 –30 min drying time. Because the nozzle sizes are not identical for the stainless steel and glass nozzles, it is difficult to predict if different substrate surfaces play a significant role. For the smallest stainless steel nozzle (1.5 mm), it was very difficult to draw the fluid into the nozzle. The SB setting was set to the maximum (10), and the 370 mg/mL BSA formulation still dried very quickly (approximately 5 min) due to the formation of a liquid plug. Overall, despite the difference in surface properties, the glass nozzle can still serve as valuable tool for predicting the optimal conditions for filling syringes from standard stainless steel nozzles. Conclusions This study identified key parameters that influence the drying rate of HC protein formulations and extend the interruption time to slow down the rate of nozzle clogging. The use of glass nozzles in a bench-top syringe filling unit offers an effective tool that enables the understanding and optimization of the function and performance of different fill parameters. Increasing the nozzle size and decreasing the water evaporation rate under appropriate environmental conditions can also effectively alleviate the tendency of nozzles to clog. More importantly, the role of the SB setting in slowing down nozzle blockage due to HC formulation drying was clearly demonstrated. A small range of optimal SB settings must be identified during filling process development. SB performance was also affected by fluid viscosity, particularly for fluids of ⬎10 cP. The effect of the substrate on SB appeared to be weak, and so observations made with glass nozzles could be applied to standard stainless steel nozzles. 162
Glass Nozzle
2.5 mm 2 90
1.8 mm
2.4 mm
4 2 20–30 ⬎60* 40.3 ⫾ 2.4
Acknowledgments We thank the support from Jacek Guzowski and Aaron Hubbard of Genentech and Ben Jones of Volo Technologies, Inc. for bench-top filling unit installation and qualification. We are also indebted to Dr. Mahmoud Ameri for his assistance in contact angle measurement. Conflict of Interest Declaration The authors declare that they have no competing interests. References 1. Romacker, M.; Schoenknecht, T.; Forster, R. The rise of prefilled syringes from niche product to primary container of choice: a short history. ONdrugDelivery 2008, 4, 4 –5. Available at http:// www.ondrugdelivery.com/publications/prefilled_ syringes_2008.pdf; accessed March 5, 2013. 2. Stockwin, L. H.; Holmes, S. Antibodies as therapeutic agents: vive la renaissance! Expert Opin. Biol. Ther. 2003, 3 (7), 1133–1152. 3. Shire, S. J.; Shahrokh, Z.; Liu, J. Challenges in the development of high protein concentration formulations. J. Pharm. Sci. 2004, 93 (6), 1390 –1402. 4. Overcashier, D. E.; Chan, E. K.; Hsu, C. C. Technical considerations in the development of prefilled syringes for protein products. Am. Pharm. Rev. 2006, 9(7), 77– 83. 5. Chana, E.; Maa, Y.-F.; Overcashier, D. E.; Manufacturing consideration in developing a prefilled syringe – investigating the effect of headspace pressure. Pharm. Rev. 2012, 15(3), 1–5. PDA Journal of Pharmaceutical Science and Technology
Downloaded from journal.pda.org on July 18, 2015
6. Kanai, S.; Liu, J.; Patapoff, T. W. Shire, S. J. Reversible self-association of a concentrated monoclonal antibody solution mediated by FabFab interactions that impact solution viscosity. J. Pharm. Sci. 2005, 97(10), 4219 – 4227. 7. Alford, J. R.; Kendrick, B. S.; Carpenter, J. F.; Randolph, T. W. High concentration formulations of recombinant human interleukin-1 receptor antagonist: II. Aggregation kinetics. J. Pharm. Sci. 2008, 97(8), 3005–3021. 8. Salinas, B. A.; Sathish, H. A.; Bishop, S. M.; Harn, N.; Carpenter, J. F.; Randolph, T. W. Understanding and modulating opalescence and viscosity in a monoclonal antibody formulation. J. Pharm. Sci. 2010, 99(1), 82–93. 9. Sukumar, M.; Doyle, B. L.; Combs, J. L.; Pekar, A. H. Opalescent appearance of an IgG1 antibody
Vol. 68, No. 2, March–April 2014
at high concentrations and its relationship to noncovalent association. Pharm. Res. 2004, 21(7), 1087–1093. 10. Tyagi, A. K.; Randolph, T. W.; Dong, A.; Maloney, K. M.; Hitscherich Jr., C.; Carpenter, J . F. IgG particle formation during pump operation: A case study of heterogeneous nucleation on stainless steel nanoparticles. J. Pharm. Sci. 2009, 98(1), 94 –104. 11. Thomas, C. R.; Geer, D. Effects of shear on proteins in solution. Biotechnology Lett. 2010, 33(3), 443– 456. 12. Deegan, R. D.; Bakajin, O.; Dupont, T. F.; Huber, G.; Nagek, S. R.; Witten, T. A. Capillary flow as the cause of ring stains from dried liquid drops. Nature 1997, 389(6653), 827– 829.
163
Downloaded from journal.pda.org on July 18, 2015
An Authorized User of the electronic PDA Journal of Pharmaceutical Science and Technology (the PDA Journal) is a PDA Member in good standing. Authorized Users are permitted to do the following: ·Search and view the content of the PDA Journal ·Download a single article for the individual use of an Authorized User ·Assemble and distribute links that point to the PDA Journal ·Print individual articles from the PDA Journal for the individual use of an Authorized User ·Make a reasonable number of photocopies of a printed article for the individual use of an Authorized User or for the use by or distribution to other Authorized Users Authorized Users are not permitted to do the following: ·Except as mentioned above, allow anyone other than an Authorized User to use or access the PDA Journal · Display or otherwise make any information from the PDA Journal available to anyone other than an Authorized User ·Post articles from the PDA Journal on Web sites, either available on the Internet or an Intranet, or in any form of online publications ·Transmit electronically, via e-mail or any other file transfer protocols, any portion of the PDA Journal ·Create a searchable archive of any portion of the PDA Journal ·Use robots or intelligent agents to access, search and/or systematically download any portion of the PDA Journal ·Sell, re-sell, rent, lease, license, sublicense, assign or otherwise transfer the use of the PDA Journal or its content ·Use or copy the PDA Journal for document delivery, fee-for-service use, or bulk reproduction or distribution of materials in any form, or any substantially similar commercial purpose ·Alter, modify, repackage or adapt any portion of the PDA Journal ·Make any edits or derivative works with respect to any portion of the PDA Journal including any text or graphics ·Delete or remove in any form or format, including on a printed article or photocopy, any copyright information or notice contained in the PDA Journal
Filling of High-Concentration Monoclonal Antibody Formulations into Pre-Filled Syringes: Filling Parameter Investigation and Optimization Wendy Shieu, Sarah A. Torhan, Edwin Chan, et al.
PDA J Pharm Sci and Tech 2014, 68 153-163 Access the most recent version at doi:10.5731/pdajpst.2014.00973
Downloaded from journal.pda.org on July 18, 2015
RESEARCH
Filling of High-Concentration Monoclonal Antibody Formulations into Pre-Filled Syringes: Filling Parameter Investigation and Optimization WENDY SHIEU, SARAH A. TORHAN, EDWIN CHAN, AARON HUBBARD, BENSON GIKANGA, OLIVER B. STAUCH, and YUH-FUN MAA* Pharmaceutical Processing and Technology Development, Genentech, a member of the Roche Group, South San Francisco, CA ©PDA, Inc. 2014 ABSTRACT: Syringe filling, especially the filling of high-concentration/viscosity monoclonal antibody formulations, is a complex process that has not been widely published in literature. This study sought to increase the body of knowledge for syringe filling by analyzing and optimizing the filling process from the perspective of a fluid’s physical properties (e.g., viscosity, concentration, surface tension). A bench-top filling unit, comprising a peristaltic pump unit and a filling nozzle integrated with a linear actuator, was utilized; glass nozzles were employed to visualize liquid flow inside the nozzle with a high-speed camera. The desired outcome of process optimization was to establish a clean filling cycle (e.g., absence of splashes, bubbles, and foaming during filling and absence of dripping from the fill nozzle post-fill) and minimize the risk of nozzle clogging during nozzle idle time due to formulation drying at or near the nozzle tip. The key process variables were determined to be nozzle size, airflow around the nozzle tip, pump suck-back (SB)/reversing, fluid viscosity, and protein concentration, while pump velocity, acceleration, and fluid/nozzle interphase properties were determined to be relatively weak parameters. The SB parameter played an especially critical role in nozzle clogging. This study shows that an appropriate combination of optimal SB setting, nozzle size, and airflow conditions could effectively extend nozzle idle time in a large-scale filling facility and environment. KEYWORDS: Prefilled syringe, High-concentration monoclonal antibody, Peristaltic pump, Suck-back, Formulation drying. LAY ABSTRACT: Syringe filling can be considered a well-established manufacturing process and has been implemented by numerous contract manufacturing organizations and biopharmaceutical companies. However, its technical details and associated critical process parameters are rarely published. The information on high-concentration/ viscosity formulation filling is particularly lacking. The purpose of this study is three-fold: (1) to reveal design details of a bench-top syringe filling unit; (2) to identify and optimize critical process parameters; (3) to apply the learning to practical filling operation. The outcomes of this study will benefit scientists and engineers who develop pre-filled syringe products by providing a better understanding of HC formulation filling principles and challenges.
Introduction Over the past decade, pre-filled syringes (PFSs) have become widely used in pharmaceutical industries and are now one of the primary containers of choice for parenteral drug delivery, especially for subcutaneous
*Corresponding Author: Genentech, a member of the Roche Group, 1 DNA Way South San Francisco, CA 94080. Telephone: 650-225-3499; Fax: 650-742-1504; email: [email protected] doi: 10.5731/pdajpst.2014.00973
Vol. 68, No. 2, March–April 2014
and intramuscular administration (1). It is particularly true for high-dose monoclonal antibody (mAb) products, which often require several milligrams of protein per kilogram body weight. The subcutaneous route of administration is preferred because it enables home administration for patients with chronic conditions, but generally restricts injection volume and time. Thus, a high-concentration (HC) mAb formulation (e.g., ⱖ100 mg/mL) is often required (2, 3). The challenge of a HC mAb formulation in the PFS product is well known—its high fluid viscosity may make syringes less desirable to use (e.g., high injection force) and formulation stability is more difficult 153
Downloaded from journal.pda.org on July 18, 2015
to maintain (3–9). The impact of HC and high-viscosity formulations on processing and manufacturing, however, is not as well documented, despite the fact that manufacturing a PFS product is a complex process that includes liquid formulation preparation (thawing, compounding, sterile filtration, etc.), component assembly (syringe, stopper, needle, and needle shield), syringe filling, stopper placement, labeling, and packaging (4, 5). Syringe filling is a complicated, but mature, operation; highly automated filling facilities have been established for many years to fill syringes while meeting strict fill weight specifications at a production rate of ⬎10,000 units/h. Yet, for engineers and scientists who are interested in learning or developing a PFS filling process, publications or disclosures detailing the syringe filling operation are scarce. Particularly lacking is published information regarding filling HC mAb formulations. This study aimed to increase the body of knowledge on HC mAb formulation syringe filling by providing insight into filling process optimization. When filling syringes using a pump mechanism, there are two sets of filling parameters: filling nozzle movement and liquid flow generated by the pump. The motion of the nozzle and the flow of the liquid must be aligned to deliver a clean filling profile, where the nozzle retracts and maintains a constant distance from the liquid level, thereby having no contact with the fluid/liquid (too close) or causing undesired splashing/ foaming/bubbles (too far apart). In addition, liquid dripping at the end of the fill should be minimized to enhance fill weight (volume) accuracy and reduce the risk of wet stoppers. When developing the PFS filling process, process engineers often use a fixed nozzle movement profile and adjust the pump parameters to complete the fill cycle. This is particularly true when the nozzle motion is driven by a mechanical cam with pre-set nozzle movement parameters controlled by a single variable—the main drive speed. Thus, this study focused only on investigating the parameters of the peristaltic pump, which is considered a standard pumping mechanism for biopharmaceutical formulations because of its mild stress and closed-system feature compared with the piston pump (10, 11). Modern peristaltic pumps control and deliver liquid through three variables: acceleration, velocity, and reversing (also called suck-back [SB] or back-suction). Acceleration and velocity determine the rate of the liquid filled into the syringe, while SB offers a particular func154
tion—withdrawing the liquid to minimize dripping at the end of each fill. Sporadic dripping may reduce fill weight accuracy. This study also delineated another function of SB, unique to HC protein formulations, in decreasing the probability of nozzle clogging due to formulation drying at the tip of the nozzle during filling interruptions (or idle time). The impact of SB in relation to several process variables on formulation drying was investigated in order to prolong filling interruption times for fluids of varied viscosity, including a model protein and a mAb formulation. The prolonged interruption time derived from this study may provide increased processing flexibility and may facilitate risk mitigation at filling sites. Materials and Methods All experiments in this study used a bovine serum albumin (BSA, at 370 mg/mL) or a mAb (mAb A at 180 mg/mL). These two molecules were formulated into the same composition containing 0.04% Tween 20. Some experiments also used distilled water containing 0.04% Tween 20 for comparison. These formulations were filled into either 25 mL PETG bottles and/or 1.0 mL long, 27G 1⁄2 inch staked needle syringes using a bench-top filling system. All equipment and materials used in these studies are listed in Table I. Bench-Top Filling System The bench-top filling unit was assembled and tested by Volo Technologies, Inc. (Roseville, CA). This unit (Figure 1) integrates a Flexicon peristaltic pump system with a linear actuator to control nozzle movement. The pump system consists of a peristaltic pump and controller, while the robotic system is composed of a ROBO Cylinder® linear actuator and a Volo-integrated controller. Peristaltic Pump Control and Tubing Arrangement: The liquid formulation was delivered using a Flexicon PD12 peristaltic pump (“a” in Figure 1) whose parameters (velocity, acceleration, SB) were controlled by a Flexicon MC12 control unit (“b” in Figure 1). Two pieces of Sani-Tech® platinum-cured 1.6 mm silicone tubing ran through the pump and were Y-connected to 3.2 mm silicone tubing before and after the pump unit (see Figure 1). This tubing arrangement was maintained throughout the whole study. Robotic Movement Control: The linear actuator (“c” in Figure 1) provides the diving action for the filling nozzle. The nozzle moves up and down per PDA Journal of Pharmaceutical Science and Technology
Downloaded from journal.pda.org on July 18, 2015
Table I Equipment and Materials Used in the Study Equipment Stainless steel nozzles (1.5, 2.0, and 2.5 mm ID; 90 mm long) Glass nozzles (1.8 and 2.4 mm ID; 90 mm long) Pump controller module (micro linear actuator) ROBO Cylinder linear actuator PETG bottle Peristaltic pump and controller High-speed camera Cone and plate rheometer Contact angle meter
Materials Syringes (1.0 mL long 27G 1⁄2 inch staked needle) Platinum-cured silicone tubing (1.6 mm and 3.2 mm ID) Y-connectors and fittings Formulation buffer Bovine serum albumin (BSA) powder
Model/ Supplier/Location INOVA (OPTIMA Pharma GmbH, Germany) Pegasus Industrial Specialties Inc. (Cambridge, Ontario) Watson-Marlow (Ringsted, Denmark)/Flexicon IAI America (Torrence, CA) Thermo Fisher Scientific Inc/Nalgene Nunc International Corporation (Waltham, MA) Watson-Marlow (Ringsted, Denmark)/Flexicon Hindsight GigE, 20/20 Hindsight, Monitoring Technology Corp., (Fairfax, VA) Physica MCR 501with CP20-0.5 measuring cone, Anton Paar GmbH (Austria) Model OCA15, FDS Corp. (Garden City, NY) Supplier/Model Becton Dickinson (Swedesboro, NJ)/Hypak Type 1 glass syringes Saint-Gobain Performance Plastics (Taunton, MA)/Sani-Tech Value Plastics (Fort Collins, CO) Genentech, Inc. (South San Francisco, CA) SeraCare Life Sciences (Milford, MA)
the commands from a Volo controller (“d” in Figure 1), which combines the function of a programmable logic controller and a human–machine interface (HMI). The Volo controller is interfaced with the Flexicon controller to send signals to start and stop
pumping the liquid. To enable diverse nozzle movement patterns (dive-in position, fill position, retraction rate, acceleration/deceleration, etc.), various set points need to be inputted to the Volo HMI and Flexicon unit. Filling Operation and Experiments The whole bench-top filling unit was placed inside a horizontal laminar airflow (LAF) hood at room temperature (20 –23 °C) with airflow at a rate of 0.7 m/s from the back panel of the hood in parallel to the bench and a relative humidity of 40 –50%. The rate of the airflow was not adjustable, but the flow could be turned off.
Figure 1 Bench-top filling system consisting of (a) Flexicon peristaltic pump, (b) pump controller, (c) linear actuator, and (d) Volo controller. Vol. 68, No. 2, March–April 2014
Nozzle Movement Profile (Cycle): The nozzle movement profile was pre-set and fixed (Figure 2) in all experiments. During operation, the nozzle dives 44 mm (measured from the back of the flange) into the syringe (52 mm in total length) and begins delivering the liquid. At the same time, the nozzle retracts with an acceleration of 150 mm/s2 and a constant velocity of 70 mm/s. 155
Downloaded from journal.pda.org on July 18, 2015
sample of 70 – 80 L was loaded onto the lower measuring plate and allowed to come to thermal equilibrium at 20 °C. A solvent trap was used to prevent fluid evaporation during the measurement. The sample dynamic viscosity was measured every 10 s for 2 min using a cone with a 19.98 mm diameter and 0.509 degree angle at a shear rate of 1000/s.
Figure 2 Graphic representation of a fixed-nozzle movement cycle used in the study.
Peristaltic Pump Parameter Setting: The fill rate of the liquid was aligned with the nozzle motion by adjusting acceleration and velocity of the peristaltic pump, as monitored through a high-speed camera. Afterwards, the SB setting, from 0 to 10, of the pump was selected and optimized to minimize liquid dripping at the end of each fill cycle, again with the visual aid of a high-speed camera. Interruption (or Drying) Studies: At the end of the fill, the filling process was interrupted (idled) for a pre-determined duration. After interruption, the fill was resumed and the fill characteristics were monitored using a high-speed camera. Suck-Back (SB) Experiments Using Glass Nozzles: Glass nozzles of different sizes (inner diameters, IDs) were used to visualize (via a high-speed camera) SB action and the flow of the liquid formulation at the tip of the glass nozzle. The SB volume, V, in response to SB settings, ranging from 0 to 10, can be calculated using eq 1 where H is the measured height of the air pocket and d is the nozzle diameter: V[mL] ⫽ H[cm] ⫻ ⫻ (d[cm]/2)2
(1)
Fluid (Formulation) Characterizations Viscosity Measurement: The viscosity of a fluid was measured using a cone and plate rheometer. Each 156
Contact Angle Measurement: Static contact angle of a fluid on glass and stainless steel surfaces was determined using a contact angle meter employing an optical contact angle method called sessile drop. The substrates were cleaned with 2% clean-in-place (CIP) cleaning detergent followed by three rinses with water for injection (WFI). For static contact angle measurement, a photo snapshot is taken once a drop of the fluid (5 L) is dispensed from the syringe and laid on a clean substrate surface (glass slides or stainless steel coupons). The angle between the baseline of the drop and the tangent at the drop boundary is measured on both sides. The complete measurement was obtained by averaging the two numbers. At least five readings were recorded for each sample. Results and Discussion Fluid Characterizations Two important fluid properties, viscosity and contact angle, were measured to determine their impact on filling characteristics and formulation drying patterns. The results are summarized in Table II for three fluids: water containing 0.04% Tween 20, 18% mAb A (buffered), and 370 mg/mL BSA (buffered). These fluids covered a broad range of viscosity, 1 to 25 cP (at 20 °C). Contact angles of a fluid on a substrate represent the interactions of the fluid and the substrate surface; low contact angles suggest hydrophilic interactions, while high contact angles demonstrate hydrophobic behaviors of the interface. Two substrate materials, glass and stainless steel, were used to represent the nozzle materials used in this study. As summarized in Table II, the glass surface is more hydrophilic than the stainless steel surface for all three fluids; the 370 mg/mL BSA solution is the most hydrophobic on both glass and stainless steel surfaces. The impact of viscosity and hydrophilicity versus hydrophobicity on fluid flow behaviors inside and at the tip of the nozzle, as well as fluid drying properties, was assessed and is discussed later. PDA Journal of Pharmaceutical Science and Technology
Downloaded from journal.pda.org on July 18, 2015
Table II Summary of Fluid Characterizations and Peristaltic Pump Parameters with Optimal Filling Profile (No Dripping) of Liquid Formulations from Stainless Steel Nozzles of Different Sizes Characterization/Parameter
Water (with 0.04% Tween 20)
mAb A (180 mg/mL)
BSA (370 mg/mL)
1 53.6 ⫾ 1.0 32.6 ⫾ 1.5 200 4
9 75.2 ⫾ 1.8 25.4 ⫾ 1.1 200 50
25 86.1 ⫾ 1.8 40.3 ⫾ 2.4 200 125
Viscosity at 20 °C (cP) Contact angle on stainless steel surface (°) Contact angle on glass surface (°) Pump velocity (rpm) Pump acceleration (no unit) SB setting 1.5 mm stainless steel nozzle 2.0 mm stainless steel nozzle 2.5 mm stainless steel nozzle
Control of Peristaltic Pump Parameters for Filling Characteristic Optimization Initially, all experiments were performed with stainless steel nozzles. With a fixed nozzle movement profile (Figure 2), pump parameters (i.e., velocity, acceleration, and SB) were visually observed through a high-speed camera to optimize filling characteristics, which included maintaining a constant distance between the fluid level and the nozzle tip and minimizing splashing, foaming, and dripping. Pump velocity and acceleration are expected to dominate the rate of liquid flow, thereby affecting the relative position of the fluid and the nozzle. Splashing and foaming may occur when the distance between the fluid level and the nozzle tip is too far. Splashing at the end of the filling cycle is a particular concern because it may result in wetting the stopper after stopper placement. A wet stopper is a critical product defect and will result in filled syringe rejection. SB does not affect solution flow rate, but it exerts a reversing action at the end of each fill cycle to modulate fluid dripping. Dripping should be prevented or minimized because undesired dripping may decrease fill weight accuracy. Additionally, dripping while the nozzle is moving out of the syringe barrel may cause splashing and/or leave fluid drop(s) on the barrel, potentially leading to wet stoppers. Optimal pump parameters that minimized issues such as splashing, foaming, and dripping for the three fluids evaluated in this study are listed in Table II. It was possible to maintain a pump velocity of 200 rpm for all three fluids, but pump acceleration had to be increased with increasing fluid viscosity (4, 50, and 125 Vol. 68, No. 2, March–April 2014
1 1 1
5 2 2
10 3 2
for water, 18% mAb A, and 370 mg/mL BSA, respectively) to align with nozzle motion. The reason that pump acceleration had to be increased in relation to increased fluid viscosity could be described by the Hagen-Poiseuille equation (eq 2): Q ⫽ R4⌬P/8 L
(2)
For a fluid flowing through a pipe—in this case, the filling nozzle—the volumetric flow rate is the function of pressure drop (⌬P), fluid viscosity (), tube radius (R), and pipe length (L). For a given pipe/tubing (fixed R and L) and a nozzle motion profile at a fixed Q, filling a fluid of higher viscosity must be countered with an increasing pressure drop, which could be achieved with a higher acceleration by the pump (i.e., ⌬PA ⫽ ma, where m is the mass of the fluid, a is acceleration, and A is the surface area where the pressure is applied). SB setting (0 to10) also had to be increased with increasing fluid viscosity to minimize dripping, but it could be decreased using a larger nozzle. These observations may again be explained by the HagenPoiseuille equation (eq 2). A greater suction force (in the direction opposite to the liquid fill) is required to pull a more viscous fluid; thus, a higher SB setting is required for more viscous fluids. However, the required suction force for a larger nozzle is lower; thus, a lower SB setting can be exerted to reduce the fluid’s dripping tendency. The function of SB turned out to be more complex and was found to affect another key filling process perfor157
Downloaded from journal.pda.org on July 18, 2015
Table III Summary of Drying Times for the mAb A Formulation Due to Interruption of Filling from Glass Nozzles as a Function of Suck-Back (SB) Settings Drying Time (min) 1.8 mm Glass Nozzle SB Setting
0.7 m/s Airflow
2.4 mm Glass Nozzle 0.7 m/s Airflow
0 m/s Airflow
0 ⬍10 ⬍15 ⬃20 1 ⬍10 ⬃15 ⬃30 2 30–40 ⬃90 ⬎120* 3 ⬍10 ⬃15 ⬃30 4 ⬍10 ⬃15 ⬃45 *No drying was observed before the end of the experiment.
mance indicator—formulation drying at the fill nozzle tip during a fill interruption. Observation and Effect of Formulation Drying During filling operations, many different type of issues (e.g., misalignments and/or malfunctions of the robotic actions, particles) may trigger filling operation interruptions. During these interruptions, the nozzles remain idle for uncertain durations. Nozzle clogging has been observed at the end of a fill interruption due to the drying of the fluid at or close to the tip of the nozzle, resulting in complete or partial nozzle blockage after the fill was resumed. This drying-induced nozzle blockage is unique to HC and high-viscosity formulations containing high-molecular-weight species (polymer, proteins, mAbs, etc.) Water evaporation from such a fluid rapidly establishes a viscous film at the drying front that can easily become elastic, thicken, or solidify. To fully understand the effect of SB setting and performance, a series of experiments was performed in glass nozzles and visualized via a high-speed camera. These experiments were designed to evaluate the effect of interruption time on formulation drying as a function of SB settings, nozzle size (1.8 mm and 2.4 mm ID), and airflow rate (0 and 0.7 m/s). As summarized in Table III, nozzle clogging took place, and in many cases it occurred within 15 min of idle time. All experiments were performed under ambient conditions 158
Figure 3 Photographic images of (a) dried residual ring forming at the inner nozzle tip and growing toward the center of the nozzle, (b) thinning flow pattern of 370 mg/mL BSA formulation after 30 min idle time, (c) graphic representation of dried ring forming for a small vs large nozzle.
(20 –23 °C and 40 –50% relative humidity) in a LAF hood, where air was either turned off (without airflow) or on (with an airflow rate of 0.7 m/s). Without airflow, it took almost twice the time for the mAb A formulation to clog the nozzle than it did under the high-airflow rate, suggesting that airflow near the nozzle tip is an important parameter affecting formulation drying. It is important to note, however, that the high rate of parallel airflow experienced by the filling nozzle in the LAF may represent a worst-case environment compared with an airflow-controlled manufacturing fill line. Table III also shows that it took longer for the larger nozzle (2.4 mm) to clog compared with the smaller one (1.8 mm). Although the rate of water evaporation per unit cross-section area should be the same regardless of nozzle size, the observation of the dried residual begins along the inner surface near the nozzle tip to form a dried ring, which grew toward the center of the nozzle (Figure 3a). The formation of an inner ring PDA Journal of Pharmaceutical Science and Technology
Downloaded from journal.pda.org on July 18, 2015
was further verified by the narrowing of the liquid flow stream (Figure 3b) after the 370 mg/mL BSA formulation was idled for 30 min. The observation of dried residual formation along the inner perimeter of the nozzle tip can be related to the observation of formation of coffee stains from dried liquid drops, which is attributed to capillary flow (12). There is a prerequisite for this phenomenon to occur—the pinning of the liquid–substrate contact line. As the edge (contact line) of the coffee drop is pinned (fixed), water evaporating from the edge must be replenished by liquid from the interior via capillary flow. In the case of liquid SB inside a filling nozzle, the edge of the fluid meniscus is pinned to the nozzle inner surface where water evaporation will easily leave a dried ring. The liquid from the interior will flow to the edge to replenish water. As this process continues, the dried ring will keep growing toward the center of the nozzle to leave a smaller opening or eventually clog the nozzle. The rate of dried ring growth is comparable between the small and large nozzles and has a more significant impact on the smaller nozzle, which will clog faster than the large nozzle (Figure 3c). SB setting was observed to play an important role on the rate of formulation drying; an interruption time could be substantially extended by optimizing SB setting (i.e., a setting of 2 for the mAb A formulation, in both nozzle sizes). The effect of SB setting on formulation drying can be explained by visual observations of the physical behavior of the formulation in glass nozzles. When the mAb A formulation was sucked into a 2.4 mm glass nozzle, it displayed a unique pattern to each SB setting (0 to 4) as shown in Figure 4a. A liquid drop hung at the tip of the nozzle at a SB setting 0 (without SB). At SB setting 1, the liquid drop was withdrawn into the nozzle to show a slight concave meniscus surface at the very tip of the nozzle. The fluid receded farther into the nozzle as the SB setting was increased to 2, 3, and 4. However, beginning at SB setting 3, an air pocket (“ap” in Figure 4b) was observed to be enclosed by a liquid plug (“lp” in Figure 4b) at the nozzle tip. The size of the air pocket and the liquid plug grew larger when the SB setting was increased to 4. When testing out interruption times at different SB settings for mAb A formulation in glass nozzles, SB setting 2 was observed to have the longest interruption time with no nozzle clogging (Table III). Using the glass nozzles, it was possible to visually observe that there was neither a drop hanging from the nozzle nor a liquid plug at the inner tip of the Vol. 68, No. 2, March–April 2014
Figure 4 Photographic images of mAb A formulation sucked back into a 2.4 mm glass nozzle (a) corresponding to suck-back (SB) settings 0 – 4 under the pump setting of velocity at 200 rpm and acceleration at 50; (b) a typical SB pattern showing an air pocket (“ap”) and a liquid plug (“lp”) at the tip of the nozzle.
nozzle at SB setting 2, which was unlike the physical behavior observed with the other SB settings. From these visual observations, a hypothesis can be made that correlates SB setting to nozzle clogging. Water can evaporate quickly from the liquid drop hanging outside the nozzle tip, and the evaporation rate can be slowed down if the fluid is pulled into the nozzle because of reduced airflow and humidity concentration gradient. Intuitively, the farther the fluid retreats, the slower the drying rate. However, like draining a liquid from a container surface, quick fluid flow/removal is normally not complete and can leave a thin film along the nozzle’s inner surface. Later, this fluid film can migrate down to accumulate at the nozzle tip to form a liquid plug which, like the liquid drop at the nozzle tip, can dry quickly and clog the nozzle. Thus, SB setting is a critical parameter that affects formulation drying/nozzle clogging, and it needs to be optimized to minimize liquid dripping or 159
Downloaded from journal.pda.org on July 18, 2015
Figure 5 Suck-back (SB) patterns for (a) 1.8 mm and (b) 2.4 mm glass nozzles in relation to SB settings of 0 –10.
the formation of a liquid plug at the nozzle tip in order to effectively extend the interruption times, particularly in filling a HC or high-viscosity fluid. The SB setting is likely related to the degree of pump rotation and is dependent on the type of the peristaltic pump used. However, despite its importance, the control of SB and the relationship between SB setting and volume are rarely specified in pump manufacturers’ user manuals, and vendors hesitate to provide any information. To generate such data, glass nozzles can serve as a valuable tool, allowing visual observations and direct measurement to quantify and assess SB performance. Control of SB Performance Having determined that SB setting is a critical parameter affecting formulation drying, additional experiments were performed to assess the effect of other fill parameters on SB. Effect of Viscosity and Nozzle Size: SB was visualized as shown in Figure 5, which portrays a linear relationship between the height of both the air pocket and liquid plug and the SB setting for two different nozzle sizes. This linearity held true in both the 1.8 mm and 2.4 mm nozzles. The SB volume can be calculated using the measured SB height (eq 1). The three solutions listed in Table II were used to assess the effect of viscosity on SB volume. As summarized in Figure 6, viscosity has an obvious impact on SB volume. At the same SB setting, a fluid with higher viscosity is more difficult to be drawn into the nozzle 160
because a higher pressure must be exerted on the tubing to draw more viscous formulations through the pump. Nozzle size appears to play a less significant role, at least for less viscous fluids, such as water and mAb A formulation (⬍10 cP). However, a nozzle size– dependent difference was observed for the more viscous BSA solution (25 cP), which could not be sucked into the 1.8 mm nozzle until the SB setting reached 4. Effect of Pump Parameters—Velocity and Acceleration: The two pump parameters, velocity and acceleration, were evaluated separately to assess their impact on SB volume of the mAb A formulation. In the first experiment, pump velocity was set to 200 rpm and pump acceleration was varied in the range of 25 and 100, represented by the solid lines in Figure 7a. SB volumes were not substantially affected by the level of
Figure 6 Suck-back (SB) volume vs SB setting for 370 mg/mL BSA, mAb A, and water in 1.8 mm and 2.4 mm nozzles. PDA Journal of Pharmaceutical Science and Technology
Downloaded from journal.pda.org on July 18, 2015
0.05 and 0.08 mL for pump velocity ranging between 25 and 400 rpm. Beyond 100 rpm, the bubble volume stayed in an even narrower range of 0.07– 0.08 mL. Because the pump parameters are typically set at ⬎100 rpm, pump velocity was again demonstrated not to be an important variable for bubble volume. Similarly, a fourth experiment was conducted with the 1.8 mm glass nozzle at a fixed SB setting of 3 and a pump velocity of 200 rpm. The results are presented in Figure 7c, where the SB volume varied between 0.05 and 0.11 mL for acceleration in the range of 25–200. Bubble volume increased faster initially (at acceleration of 25–125) but basically fluctuated in a small range (0.09 – 0.11 mL) for acceleration of 125–200. Given that pump acceleration is typically set between 25 and 125 for high-viscosity formulations (using the 1.6 mm silicone tubing), pump acceleration may have a slightly more significant impact on the SB behavior than pump velocity during syringe filling. Overall, pump velocity and acceleration were shown to be relatively weak in affecting SB behavior. Formulation Drying in Metal Nozzles
Figure 7 Suck-back (SB) volume vs SB setting for 370 mg/mL BSA (a) with varying pump velocity (V, dashed lines) and acceleration (A, solid lines) settings; (b) illustration of SB volume at SB setting 3 and acceleration at 75; (c) illustration of SB volume at SB setting 3 and velocity at 200 rpm.
pump acceleration. In the second experiment, pump acceleration was fixed at 50 and pump velocity was varied in the range of 100 and 300 rpm, represented by the dotted lines in Figure 7a. Pump velocity did not affect SB volume significantly either, suggesting that pump velocity did not play a significant role in the SB function of the peristaltic pump. In the third experiment pump velocity was evaluated against air pocket (bubble) formation in the 1.8 mm nozzle at a fixed SB setting of 3 and a pump acceleration of 75. The results are shown in Figure 7b, where the SB volume fluctuated in a narrow range between Vol. 68, No. 2, March–April 2014
Glass nozzles rendered valuable visual information for the understanding of critical pump parameters. However, glass and stainless steel have different surface properties, and it can be argued that observations made with glass nozzles are not relevant to stainless steel nozzles. Thus, an experiment was designed to evaluate whether observations in the glass nozzle can be applied to stainless steel nozzles. Earlier we pointed out that the 370 mg/mL BSA formulation is the most hydrophobic on both glass and stainless steel surfaces, but we also showed a big contact angle difference between the two: 40.3 ⫾ 2.4° on glass and 86.1 ⫾ 1.8° on stainless steel. Thus, this fluid was tested on both glass and stainless steel nozzles of different sizes (see Table IV). The optimal SB setting (i.e., positive SB without forming a liquid plug or liquid drop) in glass nozzles (1.8 and 2.4 mm) were determined to be 4 and 2, respectively. Applying these optimal SB settings, the 370 mg/mL BSA solution still dried much quicker in the 1.8 mm nozzle than in the 2.4 mm nozzle (20 –30 min vs ⬎60 min). For stainless steel nozzles of three different sizes (1.5, 2.0, and 2.5 mm), the formulation drying (nozzle clogging) results again demonstrated the same trend. At a comparable nozzle size (2.4 mm glass nozzle and 2.5 mm stainless steel nozzle), the optimal SB setting and the duration of drying times 161
Downloaded from journal.pda.org on July 18, 2015
Table IV Summary of Drying Studies for 370 mg/mL BSA Formulation Sucked Back into Glass Nozzles or Stainless Steel Nozzles of Different Sizes Stainless Steel Nozzle Study Finding
1.5 mm
2.0 mm
Optimal SB setting 10 3 Drying time for nozzle clogging (min) 5 ⬎60* Contact angle (°) 86.1 ⫾ 1.8 *Drying experiment and nozzle observation did not exceed 60 min.
were identical. For the 2.0 mm stainless steel nozzle, the SB had to be increased to 4 to significantly extend the drying time to ⬎60 min, while the 1.8 mm glass nozzle had an optimal SB setting at 3 and 20 –30 min drying time. Because the nozzle sizes are not identical for the stainless steel and glass nozzles, it is difficult to predict if different substrate surfaces play a significant role. For the smallest stainless steel nozzle (1.5 mm), it was very difficult to draw the fluid into the nozzle. The SB setting was set to the maximum (10), and the 370 mg/mL BSA formulation still dried very quickly (approximately 5 min) due to the formation of a liquid plug. Overall, despite the difference in surface properties, the glass nozzle can still serve as valuable tool for predicting the optimal conditions for filling syringes from standard stainless steel nozzles. Conclusions This study identified key parameters that influence the drying rate of HC protein formulations and extend the interruption time to slow down the rate of nozzle clogging. The use of glass nozzles in a bench-top syringe filling unit offers an effective tool that enables the understanding and optimization of the function and performance of different fill parameters. Increasing the nozzle size and decreasing the water evaporation rate under appropriate environmental conditions can also effectively alleviate the tendency of nozzles to clog. More importantly, the role of the SB setting in slowing down nozzle blockage due to HC formulation drying was clearly demonstrated. A small range of optimal SB settings must be identified during filling process development. SB performance was also affected by fluid viscosity, particularly for fluids of ⬎10 cP. The effect of the substrate on SB appeared to be weak, and so observations made with glass nozzles could be applied to standard stainless steel nozzles. 162
Glass Nozzle
2.5 mm 2 90
1.8 mm
2.4 mm
4 2 20–30 ⬎60* 40.3 ⫾ 2.4
Acknowledgments We thank the support from Jacek Guzowski and Aaron Hubbard of Genentech and Ben Jones of Volo Technologies, Inc. for bench-top filling unit installation and qualification. We are also indebted to Dr. Mahmoud Ameri for his assistance in contact angle measurement. Conflict of Interest Declaration The authors declare that they have no competing interests. References 1. Romacker, M.; Schoenknecht, T.; Forster, R. The rise of prefilled syringes from niche product to primary container of choice: a short history. ONdrugDelivery 2008, 4, 4 –5. Available at http:// www.ondrugdelivery.com/publications/prefilled_ syringes_2008.pdf; accessed March 5, 2013. 2. Stockwin, L. H.; Holmes, S. Antibodies as therapeutic agents: vive la renaissance! Expert Opin. Biol. Ther. 2003, 3 (7), 1133–1152. 3. Shire, S. J.; Shahrokh, Z.; Liu, J. Challenges in the development of high protein concentration formulations. J. Pharm. Sci. 2004, 93 (6), 1390 –1402. 4. Overcashier, D. E.; Chan, E. K.; Hsu, C. C. Technical considerations in the development of prefilled syringes for protein products. Am. Pharm. Rev. 2006, 9(7), 77– 83. 5. Chana, E.; Maa, Y.-F.; Overcashier, D. E.; Manufacturing consideration in developing a prefilled syringe – investigating the effect of headspace pressure. Pharm. Rev. 2012, 15(3), 1–5. PDA Journal of Pharmaceutical Science and Technology
Downloaded from journal.pda.org on July 18, 2015
6. Kanai, S.; Liu, J.; Patapoff, T. W. Shire, S. J. Reversible self-association of a concentrated monoclonal antibody solution mediated by FabFab interactions that impact solution viscosity. J. Pharm. Sci. 2005, 97(10), 4219 – 4227. 7. Alford, J. R.; Kendrick, B. S.; Carpenter, J. F.; Randolph, T. W. High concentration formulations of recombinant human interleukin-1 receptor antagonist: II. Aggregation kinetics. J. Pharm. Sci. 2008, 97(8), 3005–3021. 8. Salinas, B. A.; Sathish, H. A.; Bishop, S. M.; Harn, N.; Carpenter, J. F.; Randolph, T. W. Understanding and modulating opalescence and viscosity in a monoclonal antibody formulation. J. Pharm. Sci. 2010, 99(1), 82–93. 9. Sukumar, M.; Doyle, B. L.; Combs, J. L.; Pekar, A. H. Opalescent appearance of an IgG1 antibody
Vol. 68, No. 2, March–April 2014
at high concentrations and its relationship to noncovalent association. Pharm. Res. 2004, 21(7), 1087–1093. 10. Tyagi, A. K.; Randolph, T. W.; Dong, A.; Maloney, K. M.; Hitscherich Jr., C.; Carpenter, J . F. IgG particle formation during pump operation: A case study of heterogeneous nucleation on stainless steel nanoparticles. J. Pharm. Sci. 2009, 98(1), 94 –104. 11. Thomas, C. R.; Geer, D. Effects of shear on proteins in solution. Biotechnology Lett. 2010, 33(3), 443– 456. 12. Deegan, R. D.; Bakajin, O.; Dupont, T. F.; Huber, G.; Nagek, S. R.; Witten, T. A. Capillary flow as the cause of ring stains from dried liquid drops. Nature 1997, 389(6653), 827– 829.
163
Downloaded from journal.pda.org on July 18, 2015
An Authorized User of the electronic PDA Journal of Pharmaceutical Science and Technology (the PDA Journal) is a PDA Member in good standing. Authorized Users are permitted to do the following: ·Search and view the content of the PDA Journal ·Download a single article for the individual use of an Authorized User ·Assemble and distribute links that point to the PDA Journal ·Print individual articles from the PDA Journal for the individual use of an Authorized User ·Make a reasonable number of photocopies of a printed article for the individual use of an Authorized User or for the use by or distribution to other Authorized Users Authorized Users are not permitted to do the following: ·Except as mentioned above, allow anyone other than an Authorized User to use or access the PDA Journal · Display or otherwise make any information from the PDA Journal available to anyone other than an Authorized User ·Post articles from the PDA Journal on Web sites, either available on the Internet or an Intranet, or in any form of online publications ·Transmit electronically, via e-mail or any other file transfer protocols, any portion of the PDA Journal ·Create a searchable archive of any portion of the PDA Journal ·Use robots or intelligent agents to access, search and/or systematically download any portion of the PDA Journal ·Sell, re-sell, rent, lease, license, sublicense, assign or otherwise transfer the use of the PDA Journal or its content ·Use or copy the PDA Journal for document delivery, fee-for-service use, or bulk reproduction or distribution of materials in any form, or any substantially similar commercial purpose ·Alter, modify, repackage or adapt any portion of the PDA Journal ·Make any edits or derivative works with respect to any portion of the PDA Journal including any text or graphics ·Delete or remove in any form or format, including on a printed article or photocopy, any copyright information or notice contained in the PDA Journal
